[electronupdate] has done a lot of LED light bulb teardowns over the years, witnessing a drive towards ever-cheaper and ever-simpler implementations, and suspects that LED light bulb design has finally reached its ultimate goal. This teardown of a recent dollar store example shows that cost-cutting has managed to shave even more off what was already looking like a market saturated with bottom-dollar design.
The electrical components inside this glowing model of cost-cutting consists of one PCB (previously-seen dollar store LED bulb examples had two), eleven LEDs, one bridge rectifier, two resistors, and a controller IC. A wirewound resistor apparently also serves as a fuse, just in case.
That’s not all. [electronupdate] goes beyond a simple teardown and has decapped the controller IC to see what lurks inside, and the result is shown here. This controller is responsible for driving the LEDs from the ~100 Volts DC that the bridge rectifier and large electrolytic cap present to it, and it’s both cheap and clever in its own way.
The top half is a big transistor for chopping the voltage and the bottom half is the simple control logic; operation is fast enough that no flicker is perceived in the LEDs, and no output smoothing cap is needed. The result, of course, is fewer components and lower cost.
Some of you may recall that back in the early days of LED lighting, bulbs that could last 100,000 hours were a hot promise. That didn’t happen for a variety of reasons and the march towards being an everyday consumable where cost was paramount continued. [electronupdate] feels they have probably reached that ultimate goal, at least until something else changes. They work, they’re cheap, and just about everything else has been successfully pried up and tossed out the door.
They work, they’re cheap, and just about everything else has been successfully pried up and tossed out the door.
Including EMC compliance at a reasonable level. They even interfere with DAB radio nowadays, well done.
The capacitive dropper bulbs are pretty low EMI.
But PF issues and a reliance on line frequency.
Stuffing a 50-ish semiconductor high-power-ish logic system and a bunch of other stuff into a small container and having it be cost-competitive with a fine tungsten wire on the promise of energy savings is pretty extraordinary. I’m wondering where the safety tipping point lies with ever cheaper and rougher products. Likely a horrific fire and non-payment of insurance because of using u”nrated equipment”.
Also: “…The electrical components inside this glowing model of cost-cutting …” I saw what you did there.
I am curious about the about the actual savings that the average LED bulb provides. Sure, LED bulbs use less electricity, and they last longer, but there must be some point where the cost of materials (LEDs, circuit board, and other components), and the electronic scrap produced after they do burn out, become more of a burden than the simplicity of a halogen bulb.
They’re plenty. That discussion was over > 10 years ago.
Here, this is a study from 2009 (when LEDs were less efficient than they are today, and production was far lower in volume):
http://seeds4green.net/sites/default/files/OSRAM_LED_LCA_Summary_November_2009.pdf
It’s admittedly been ordered by OSRAM and SIEMENS, both of which have business interest in LEDs looking good, but the numbers added up in the study seem reasonable. And 2009, LED lamps would have had to be 5 times worse than the study claims to get as bad as old-school bulbs.
Really, this discussion is more than over.
That study is a bunch of crap. LEDs last 1/10th as long as promised, created plenty of toxic waste in production and do not recycle. The march to ever cheaper crap just takes more money out of our pockets and puts more toxic waste in landfills and you all call it going green. The current generation are a bunch of low watt wonder dummies.
really? Like, you look at that study and go “oh, there’s concrete calculations using real numbers, but here’s my opinion on electronic waste which obviously tops that”? You know what, how about you actually show us at least a napkin calculation that proves your point.
luckily, I seem to be buying a different brand of LED lights than you are. Considering the ones illuminating this very room right now: They’re well beyond 2500 hr of operation, which would be “1/10 of what they promised”. Check your confirmation bias, yo!
Yep, it is a bunch of crap. It ignores the energy and materials used in the creation of the LED bulbs.
Over half of my home GE “HD” LED bulbs have failed in ~2-1/2 years. That’s 14 bulbs and counting. Perhaps that’s the difference between GE and other brands, but I suspect it’s all BS.
As far as I am concerned the spiral bulbs last much longer then incandescent and therefore are 1/10 the landfill. LED bulbs list twice as long Aspire rules. I think it’s pretty obvious that manufacturing one light bulb LED Style vs 20 or 30 incandescence saves materials coolest and energy in itself. The other grade saving it is very often forgotten is that we have to cool off the spaces incandescent bulbs are heating up vs LED bulbs
j1caesar – write for yourself. I for example live in a place where heating is on about 1/2 of the year.
So – energy efficient lamps? what is the point? ;-)
(Halogen bulbs, especially low voltage ones with proper power last very long – my bathroom lights are still original ones I installed in 2011).
That study is a bunch of crap, but for a different reason: compact fluorescents. Comparing LED bulbs with incandescent is a straw man comparison.
CF bulbs already are almost the same price as incandescent bulbs, and very low power consumption. Almost everyone is already using CF bulbs (in Europe/ U.K. at least).
CF bulbs also are (mostly) available in decent light outputs, have a reasonable longevity, and good light colour (cool, warm, or daylight, as you please).
Anecdotally, CF spotlights (G10?) are crap though. I’ve had loads die very fast, and no, not overheating as they’re not enclosed. LEDs I’ve switched in have been much better. I suspect there’s a minimum size for CFs to work well, whereas LEDs excel at small size and lower power.
LEDs also excel for adjustable colour temperature – I fitted our office with some big LED down lights. They’re not a standard consumer thing, and not cheap, but work excellently and have been extremely reliable. Cost effective for the benefit they give, but definitely not cost effective just for light output.
Compact fluorescent lights are even worse than LED lights when not properly recycled. At least LED ones do not contain mercury. (And they have electronics in them anyway).
I was thinking the same – I’ve had a good number of led bulbs ‘burn out’ (driver circuitry) due to how horribly cheap they are made. They could do so much more good for an extra dollar to have a huge lifespan and not be disposable, but then how would the companies make money if they weren’t consumable? Ugh…
It seems like having a required deposit that you get back when you return them (like aluminum cans) would reduce the incentive to sell the very cheapest bulbs.
You mean another debacle like what’s happening in California?
https://www.latimes.com/environment/story/2020-01-13/california-recycling-beverage-industry-cans-and-bottles
Most cities in California no longer recycle. Los Angeles closed their centers. Now all our recyclable metal, glass, plastics, and papers go straight to landfill. Deposits go to someone else who never has to pay them back. And these burned out lights would go right into landfill. No. We have to reject the disposable, non-repairable, non-sustainable technology and politics.
I bought DIY ring light kit from China. Their intended market was fluorescent replacement. The driver and LED are separate. I bought them thinking that they are a bit more serviceable as I got spare SMT LED chips. The driver died after 1.5 years and I replaced a cap to fixed the driver, but the LED worked for another 3-4 months before they failed. They didn’t connected the thermal pads, so multiple LED packages cracked in a cascaded failure. In a sense, they had a balanced design as neither components were over engineered over the other. :P
When my replacement die, I’ll replace it with regular LED bulbs similar to one in OP’s as they are now available in my grocery store. it was ironic that I started with CF bulbs.
Our old flourcent tube is much better then these LED
The main issue I see is the use of Edison socket. It is too small, the bulb holds LEDs and driver in small package, the heat is concentrated, so they fail prematurely. What about separating driver and LEDs. In that way driver may be quality one, I would pay several dollars (even more) if it would last for decade(s). By distancing driver and led heat is less of problem, also. LEDs can be spread on some wider surface, producing nicer light with less heat….
But who would buy bulbs in quantity then?
The socket is a big problem, as is the power grid in general. Most of the appliances in our homes now run on DC power, typically 5 volts or less. Wall warts convert our AC house current down but any time you convert electricity some is lost in the conversion, usually as heat. I think what we might see in the future are low voltage dc lines running in the house along with the current 110 and 220 ac lines. It makes more sense to have a single transformer in the home converting things down to DC rather than several wall warts. I could easily see all of the lighting on a home running on either 5 or 12 volts with the “bulbs” plugging in via a usb-like connector.
@HowardC: Please do not propagate this junk any more. Low voltage is unsuitable for distribution over more than a few meters. And it is more effort to “transform” DC for HV transmission and back. The big iron core transformers of the utility services work for decades because they do not contain any electronics.
A DC grid is no solution and even in a household a centralized DC supply rarely makes sense. Not even “USB” is always the same, think on Smartphone quickcharge with voltages up to 9V or 12V. And sometimes the voltage will be adapted and controlled by the supplied device to match the state of charge of the battery. There is no efficient way to do this centralized.
Low voltage DC distribution inside a home is a bad idea because of the energy loss and massive amounts of copper required for the cabling. 100 Watts at 12 volts needs 8 amps – going by the rule of 3 amps per square millimeter, you need 12 AGW cable and you’ll still lose almost a volt (7%) for every 10 meters of cable.
DC power distribution inside a home with high current and/or high voltage is a bad idea for fire safety reasons as well, because a DC arc in a switch or a frayed wire does not distinguish itself when the direction of the current changes, because it doesn’t change. There’s a very good reason why switches and relays are rated for much less DC than AC amps.
“Extinguish”, not “distinguish”.
Anyways, everything with DC has to be built extra robust because AC has the neat property of cutting out every 1/120th of a second, which means switches and sockets can’t sustain an electric arc when they’re being separated and don’t easily weld together.
I had wondered the same thing. And my thought was why doesn’t the light switch be the AC/DC transformer, but after reading the comments about DC and voltage loss, etc. that would be an issue.
So may be the light fixtures should have the electronic? And come out with a new plug style to match fixtures and bulbs.
Edible bulbs.
Good plan, let’s get up to speed on that CRISPR stuff and splice glow-worms with onions.
Absolutely right. I have bulbs lying in dozens which started flicking between 2 and 12 months. Obviously cheap components are being used.
All these people talking about failed LED bulbs. Please tell me what brands they are so I can continue to avoid them. In 7 years time I’ve yet to have a single LED bulb fail on me. A few of them are even on all day and night.
For home use type bulbs the glass is often coated to defuse and soften the light, more often than clear glass anyway, and that makes for some considerable recycling constraints that are shared between both types of bulb. So lasting longer alone helps to reduce that waste.
I’d think the PCB/component scrap compared against comparable scrap from today’s disposable minded consumers is a drop in the bucket, one of a larger issue that should be addressed in whole.
Compared to home incandescents, LED are less energy to use and now less energy to make, so a win all around there.
You are probably quite right about halogen, but those really are a minority usage compared to the home use market. But I’d be curious too, LED isn’t a full replacement for just anything, and halogen is one of many minority cases, so all of those added up (especially from the industrial market) are probably a larger percentage than expected.
Why would an LED bulb require less energy to make than an incandescent bulb?
I would guess not having to melt and blow all that the glass helps. The burn-in period would be automatically cheaper, if they run through one.
sigh, my grammar, the editing forbearance. Such Engrish.
While I do like (and use)the LED bulbs at least the ones that seem to be made decently, I like halogen better. I have a couple large 500w halogen lights I use occasionally and some halogen flood bulbs & the 72w halogen bulbs that look like normal Tungsten bulbs in my workshop where I need lots of bright lightning. I still use a few CFL bulbs & have a double 22 & 32 watt fluorescent T9 fixture in my kitchen, which those bulbs last a very long time and you don’t have to worry about underrated electronics failing.
Unfortunately especially the large lights like 500W also produce excessive heat what makes them quite unsuitable as work lights except in cold climate.
When it’s warmer you often already sweat from the work and do not need any additional heating.
I prefer the halogen lamps also.
The manufacturer will not give them to you for free. The cost of the material are included in the purchase price. So your fears are without any reason.
Bulbs based on led filament diodes seem like the finale form… No passives, no ic, no extra components to fail or contribute to inefficiency.
If there’s no other components than the filaments, then they’re the worst possible type because they blink with the AC frequency. They’re just doubled up so half the diodes conduct in the other direction, making a 120 Hz flicker.
I don’t know a lot about the filament types technically, but I have some, they continue to glow for several seconds after flipping off the switch, and I have never perceived a flicker from them. That said, they also use more energy than the regular LED models.
The problem with filament type is license costs. Some university invented them and is currently suing the sellers, not the makers.
Surprise surprise.
Why are they going after the sellers instead of the manufacturers? Are the manufacturers in China or some other country where they can’t be sued?
The thing is, the university didn’t invent filament LEDs. What they actually invented is some overly complicated and impractical redesign of the old-fashioned 5mm resin dome LED that was meant to use total internal reflection to direct the light. It has nothing in common with filament LEDs, which of course emit light in all directions and have a totally different structure – but they managed to get a patent with such ludicrously overbroad claims that they could use it to sue sellers of filament LEDs anyway.
That’s a curiously long phosphor then.
And the phosphor types typically use less energy than the regular cheap “white” LEDs. Are you sure you aren’t confusing it with a CFL bulb?
Yep, if they even bother to to run LED’s both directions. The name-brand (Phillips) Christmas tree lights I had didn’t even bother to do this, so the entire string was only on for part of a half cycle (makes it cheap and easy fort them, not to mention they can advertise even lower avg wattage while they’re at it…) – Maybe I’m more sensitive than most, but the flicker drove me nuts enough to put a bridge rectifier inline I had laying around to at least keep the on-time up. Bridge in the first string, and series connected strings are good to go also. Helped a ton. – Built to the bare minimum cost…
Good point. Checked led filament style vs our sole incandescent bulb using 960 fps phone camera. Filament led bulbs were barely perceptible and I had to look at a couple spots to actually just notice a flicker. Incandescent bulb performed much worse with a noticeable flicker right off the bat. Something like 20% brightness variation.
Win win for me so far :)
Put a blue filter in front of the camera and check again. The yellow phosphor coating of the filament glows steadily while the blue or near-UV diodes inside the filaments blink like crazy.
Nope. Several years ago I measured the decay time of the yellow vs blue light in high power “white” LEDs, when I was building a high-speed LED flash.
The blue follows the fall time of the electrical current, as you might (naively but correctly) expect. 100 nanoseconds in my measurement case. The yellow phosphor light decays with a 2 microsecond time constant.
Unfortunately, diodes are not linear. They don’t really start conducting until very close to the knee voltage (~3v) and then they present an AC resistance of almost zero ohms. So the EMI from LED only lamps will be truly horrific.
> ~100 Volts DC that the bridge rectifier and large electrolytic cap present to it
It is a heck lot more than 100V DC. If you don’t understand that peak voltage is sqrt(2) (1.414) of your usual 120V Sine wave, well you might want to do more research.
https://electronics.stackexchange.com/questions/250887/how-to-calculate-voltage-after-rectifier
These bulbs (by design, because it improves power factor) do not have enough filter capacitance to catch the peak voltage of the mains. In operation the voltage on the DC bus varies from the 160V peak down to just above the dropout voltage of the LEDs (about 30v usually), with the average voltage usually about 100v depending on the design.
Which also means the filter capacitors take a real beating and the 22.8 year lifespan as advertised is complete bunk. 2- 5 years in reality.
Ripple current rating seems to be the most often neglected characteristic of capacitors, and at the same time possibly the most important determining factor for a caps expected life.
For anyone unfamiliar with reliability analysis, it’s a good idea to become familiar at least with the concepts to improve your designs. Often the difference between adequate and abysmal service life is not very costly. See MIL-HDBK-217 if you want to learn.
That handbook is old and not very accurate any more, at least according to https://www.designnews.com/electronics-test/end-near-mil-hdbk-217-and-other-outdated-handbooks/138268218059056
Those caps also fail from being heated so much. I received a bunch of free LED bulbs from the power company and all the 60w equivs burned out in a year. 22.8 years on paper sound means the ambient electronics temp is 20c. That isn’t happening.
People say oh buy the good ones, sure $10 for a light bulb and then I have to stash the receipt away for when it does fail. So then I pay to send back the failed one? This is sounding better and better.
Another common trick is to declare the brightness of the bulb at the nominal current/wattage of the diode before applying the de-rating curve for temperature or current, so you can have a nominal 8 Watts worth of diodes producing 960 lm of light (60 Watt equivalent) on paper, but actually the diodes are run at a lower than nominal current to stop them from burning out (yet they’re still overheating) which drops the light output by a quarter to a third of the nominal figure.
What counts as a 60 Watt “equivalent” is also a gradually diminishing number, and it’s currently going somewhere around 700-800 lumens to cover for the fact that the LEDs don’t put out as much light as claimed. Meanwhile, I still have incandescent bulbs that are rated at 75 Watts and 1300 Lumens (17.3 lm/W) meaning a 60 Watt bulb should put out over 1000 lm. The trick there is to compare the LED against frosted and/or tinted glass bulbs rather than clear bulbs to get a more favorable comparison.
They get away with this because it’s an industry standard to test diodes too fast to account for self-heating. Otherwise you just couldn’t test them, so it’s the responsibility of the bulb manufacturer to apply the proper de-rating for the actual conditions and print this information on the box. As a consequence, you get bulbs that test to be within specs, but dim out almost immediately in use and become permanently dimmed in a few hundred hours of use.
If you live anywhere near an IKEA, try their LED bulbs. I’ve installed hundreds of IKEA LED bulbs in 20 houses, and I have 6 years of maintenance records on them. The number of failures so far? Zero.
My current house has IKEA bulbs in exterior fixtures which are powered on 24/7. In the summer, ambient temperatures here exceed 120ºF. And when sun hits the fixtures, they get much hotter. Yet the bulbs continue to shine. They’re 6 years old.
I’ve tried many brands of LED bulbs, but none have come close to IKEA’s reliability record.
(For comparison, there are 4 dead “Bazz” brand LED bulbs sitting next to my keyboard right now. They’re rated for 25,000 hours. None of them lasted more than 40 hours in service.)
I have a feeling that the capacitor is actually connected after the chip. Each cycle, the chip turns on at zero crossing. The cap charges when the input voltage is higher than the voltage it holds until the chip stops conducting once the output current reached a preset limit.
This is similar to the usual driver using an inductor to maintain current except it is a cap holding voltage. It is not as good at maintaining LED current, so the capacitor value has to be bigger. Since the cap voltage is not going to be more than 40V or so, they use a 100V cap leading the author to *think* that the rectified voltage is ~100V.
I missed something like capacitive isolation (Cuk or something like that). It’s a little safer and require more or less the same parts
My guess: either that IC is a actually a linear current regulator, or that wirewound resistor is actually an inductor. Most of the designs I’ve pulled apart have used linear current regulation into a fairly high voltage (~100V) string of LEDs.
As a previous commentator pointed out, the capacitor filter in these bulbs is only enough to keep the voltage above 30 V over the full cycle and the ripple on top is more than 100 Volts. The idea is that the circuit draws current whenever the voltage is above the 30 Volt LED threshold, so when the LED threshold is low the current is on for a greater portion of the cycle and the power factor of the circuit gets closer to 1.
This means that any linear regulator would have to drop the extra voltage at roughly 70 mA of current and waste about half the power output of the whole bulb, so instead the circuit is running in a switched mode. When the rectified voltage dips close to 30 Volts, the diodes are barely conducting, and when the voltage rises to 160 Volts or so, the chopper circuit uses the wire-wound resistor as a ballast to limit the current to around 70 mA. This causes the bulb brightness to dip down 120 times a second, which gets gradually worse when the capacitor starts to croak from the heat and the extreme current ripple.
Linear regulation works if the rectifier is built to keep a low ripple and high voltage, so the LED threshold can be raised close to the line voltage. This means having a large filtering capacitor (more expensive), and a poor power factor since the bulb draws current only when the line voltage is above the capacitor voltage. Poor power factor means more EMI and not passing regulations in some markets.
The resistor does double duty as an in-rush current limiter when the light is turned on.
I’d certainly buy that *if* the resistor were of significant value, but if the color code (brown-green-black) is to be believed, it’s all of 15 ohms. Assume a half-watt resistor (it’s almost certainly less) with a factor of two peak power rating and you’d never want to drop more than 3.9V across it which is not much ballasting at all. It would be interesting to know the actual value of the resistor.
It’s a wire-wound resistor. It acts as an inductor at high switching frequencies, and limits the in-rush current to the capacitor at switch-on to a maximum of 10 amps.
That resistor looks like a film resistor not a wirewound type. It acts as an inrush current limiter and fuse. Old fashioned incandescent bulbs could draw a startup current of ten times running current and on failure that flash often seen could run to 100 times normal draw.
The blue body color normally means “Flame Proof” meaning it will not catch on fire not that it won’t do a nice bit of spark when it fails due to over current.
Decent electrolytic capacitors are normally rated for 2,000 hours of life, meaning their capacitance value will drop 10-20% by that time. It drops even more when there is high ripple current. There are others rated for 1,000 to 10,000 hours.
There’s a picture of it where they filed the side off, and you can see the wires.
what I’d like to know, is what’s the best build on the market? We all have bought the cheap LED bulbs and had a few of them burn out after a year… so what I want is to know what bulbs have the internal components to actually last 100,000 hours? Ignore the LED rating, give me MTBF for the rest of the crap.
The original Cree LED bulbs from 2013 were probably the best. I bought a few back then (I think for $10 each from Home Depot) and all of them are still running. One’s been my porch light for the last 7 years, running at least 25k hours. But no one wants to pay $10 for a bulb if they can pay $2. I tried digging around on Google to find the MTBF for them but the best I came up with was they had a 10-year warranty.
>But no one wants to pay $10 for a bulb if they can pay $2
They will if they’ll be paying less money in the long run.
Unfortunately most people don’t really think that way. I’ll admit I’m one of them sometimes “oh, I’ll only use this tool a few times, surely I don’t have to pay more than this, it’ll be fine”… *sigh*
It only make sense if/when the $10 light blue last at least 5X longer than the $2 one or bulb is located at some hard to reach location for frequent replacement. In reality, t the high price bulb would be hard to find and require special order because the stores sell bulb priced between 0.8X to 1.5X that consumer want.
Or worse, someone would sell a bulb for 5 times the price, advertising that it is a “premium long-life LED” and skip town before they start burning out.
Don’t have to skip town, just do the usual, “Warranty only serviced if shipped in original packaging through a bonded tracked courier” and keep the sticker lower than the $15 to $20 that costs.
Ima surprise “Noma” sometime soon though, send them a dozen in one package.
Or please include the *original* sales receipt that was printed on thermal paper. :p
A lot of products sold here that would cost a large % of its retail cost just to ship them back to their manufacturer with an insured international postage. The $50 PSU with 5 years warranty isn’t worth paying $20 to ship across the border.
I’d like to find bulbs that use a Cree LED or Cree Style because I have a flashlight & bike light that use a 26850 and 18650 battery. The flashlight is just a little larger than the battery so its a little small but that thing is super BRIGHT! It would put anything with a krypton or halogen flashlight bulb to shame even brighter than some older spotlights.
The cheapest way of getting LEDs that last a very long time is to buy more power rating than you need and then turning them down, either with a dimmer for dimmable LEDs or by modifying the driver circuit.
And installing surge protectors. They really don’t like poor power quality.
in my experience, none of them. of course i have a couple led bulbs in my bedroom that are half way there. the 100000 hours mark also assumes ideal conditions, i think i bought a dozen or so bulbs (i forget the brand) and maybe 3 of them still work. many of the others were installed in enclosed fixtures which i believe is what killed them prematurely. the fixture in my room is open and allows for proper ventilation. you would think with the higher efficiency, lower power consumption and lower heat output, these fixtures would be able to handle those bulbs. but semiconductors are significantly more sensitive to heat than filaments. this was also a huge problem with cf bulbs as well, the electronics would die long before the florescent tubes in most cases.
Compton Lamps of Bradford, England. Best reasonably priced lamps in U.K., though I am sure that they will export, also the oldest lamp manufacturer in the world. Founded in 1878.
I only buy Osram or Philips. Maybe I’m just lucky, but from all the lighting I have installed at home right now only a single bulb died before the end of 2-year guarantee. Earlier I used to buy the cheapest ones and got fed up with dealing with returns or shopping for new ones every couple of months.
Also like Osram, had bad luck with Philips.
The longest-lasting LEDs in our house right now are some Toshiba lamps that were meant to be in-built in ceilings, and thus have absolutely massive cooling fins on them. Got them cheap at a surplusser, but let me just say that it’s all about the cooling. (Which I guess we already know…)
Ain’t that typical though, finding superb products at closeout or surplus, then by the time you figure out how good they are (Because you only bought in sample quantity, because it was a risk.) they’re all gone. I’m always all “Why come I never knowed ’bout these when they were at retail???” I figure marketing or distribution fail must be part of the story about why they ended up there.
I am using a $3 Chinese LED bulb from aliexpress. There are no chips in it. There is only a film cap (cap drop) + LED on PCB. I put in the washroom fixture that apartment have installed. It sees a lot of on/off cycles. So far it passed the 2 years mark which is how long a decent CF lamp .
I’d guess the primary point of failure, if not something mechanical, is going to be the electrolytic cap. I doubt they use the solid electrolyte type, so 20k hours of operation seems unlikely.
i kind of think the we we do led lighting is all wrong. trying to put the current regulator and leds in a lightbulb shaped enclosure, and then putting those bulbs in fixtures that are not designed to reject waste heat effectively at a level acceptable for semiconductors.
what i think we need is special led fixtures. which would contain a high quality current regulator and be designed with adequate cooling in mind. the only consumables would be the led modules themselves, a small pcb containing just the consumable parts, like the leds themselves and anything else that can wear out. to keep the cost down they would plug into an edge connect socket, with the other side of the pcb being screwed into a heat sink. much of the copper on the pcb would actually be used for heat removal so you could just have a metal on metal contact with the heat sink. this would drive the module cost down a lot being just pcb and leds.
cooling could be either active or passive. passive designs might have a heat pipe take the heat to a convection zone where heat rising up from the fins draws in cool air from underneath. this would all be a permanent part of the fixture. larger fixtures might be cooled by low noise fans (also considered a consumable). the current regulators would be sized to the fixture, so a multi led fixture might have a better rated regulator. and active cooled designs the regulator would also have provisions for a fan connector. the current regulators themselves might also be modular so a failure there would not doom the whole fixture.
First you need to standardize the LEDs so that they have the same voltage and/or current specifications, and de-rating for temperature. Then you can design a standard fixture.
The Edison socket is actually pretty good for thermal management. It’s just that keeping things cooler than semiconductors and electrolytic capacitors would like is difficult all around – and it’s unlikely you can actually design a better fixture without compromising other facets of the design.
The problem is few people want to replace any of their share of the millions upon millions of Edison socket lighting fixtures. They want something “screw and play” so they can simply swap the bulb and forget about it.
This is how my lighting is done in my house here in Japan.
I have Panasonic brand ‘down light’ fittings throughout (ES and BC globes aren’t as common here these days), and the power supply is seperate to the light fitting itself with the actual LED module in the light fitting replaceable.
You don’t even need to remove anything from the ceiling to replace the LED module.
I opened one of the power supplies up and was plesantly surprised to find United Chemicon caps being used, ones with a high ripple current rating too.
I have an LED light panel like that where the PS is a separate big ballast. I think where all the problems come from is squeezing a lot into a little space.
Can we get a model number for the fixture or at least led module?
Heat is the enemy of electrolytic capacitors. Keep them cool and the electrolyte doesn’t evaporate. They have ratings for temperature and life, and the lamp manufacturer should buy the appropriate capacitors and not overstress them.
That’s not how you make money in the _consumables_ market. Either the product is designed to break, or it’s otherwise poor in quality to the point that the consumers will be enticed to replace it with the better model you introduce a year later. Or both.
There is an incandescent bulb that has been on for over 100 years. It’s not that incandescent bulbs can’t last, they aren’t made to.
The same thing is happening with the Edison socket based replacement bulbs. If you want longevity, the fixtures with the leds and power supply built in may be better.
The manufacturer has complete control over their product, including cooling. The downside is that the whole fixture will need replacement when the leds eventually fail.
That bulb in that San Francisco firehouse has lasted so long because for the past 50+ years it has been fed a low electron diet of just barely enough power to make it visibly glow.
that’s just survivorship bias. you have to realize that all of its siblings have long since died and were sentenced to the landfill long before any of us were born.
The problem is that you are not always getting a better product for paying more. You can easily pay more to get a “better product” and get the same as or worse than the cheaper product. You can’t really open and see the circuit of the lamp before buying it, so you just end up trying your luck most of the time.
These definitely aren’t better.
The light output is about 70-80% compared to the Philips bulb they used for a benchmark, which corresponds well with the luminous efficacy of these diodes. The Philips bulb is specified at 90+ lm/W and these bulbs achieve about 60-75 lm/W. The second problem is the poor CRI, again about 75 vs. 90+ compared to the L-prize bulb.
The Philips bulb that initially cost more than $20 is almost as good as an incandescent bulb, while these dollar store LEDs are worse than fluorescent lights both in terms of efficiency and light quality.
Better than what? I didn’t compare these to any other and in any way say these were good bulbs. What I’m saying is that most of the time you can’t know the quality of a product (not talking about LED bulbs, any product) without buying and tearing it apart. Sometimes you can find someone who already did it for you, but most of the time you don’t.
While we are at it, Phillips bulbs are nice and all, but really not easy to find around here (only those LED filament ones) and prohibitively expensive when found. It’s not a matter of $2 to $10, it’s $2 to $50. I would gladly pay $20 for a bulb with 90CRI, but I can’t even find any better than 80 around here.
Unfortunately we are flooded with Chinese import trash and it becomes harder and harder to find good products.
Better than the cheap products. All the cheap LED offerings are just awful – only fit for a porch light.
I hear you. I have some commercial grade Osram-LED-Spots in use (in 2015 they were about $20 each) which sport ‘made in Ukraine’ on the case. They work REALLY great; kind of the best color rendition and light output you can imagine. Never had one fail until now. Steep price but worth every penny.
Of course they are discontinued now and replaced by the usual cheap Chinese sh*t.
https://www.energy.gov/eere/ssl/led-lighting-facts
DOE LED Lighting Facts® program
https://indiecinemaacademy.com/complete-led-color-database-cri-tlci-cqs-tm30-15/
Complete LED Color Rendering Database of 2018: CRI, TLCI …
What good is that second link? It’s all stage and set lighting.
To the author/editor: The parts list doesn’t list a capacitor, which the very next paragraph mentions. Since the parts list is central to the story, you may want to fix that.
The ultimate would be a single chip, wirelessly powered, solution with dimming plus color control, and so networked IoT capabilities. Not sure if we will ever see that but my intuition says the laws of physics would allow it, even if you’d need a lot of them as their individual power output would be low. Imagine that, an array of light-motes scattered on any surface, all of them individually addressable and controllable.
Why?
I don’t usually change the colour of my lights often. I don’t need to be able to put them everywhere.
Once you have got to the point you are talking about, point a camera at them and you can work out which address is where, and then every surface becomes a display… And before long we will have adverts everywhere.
Aesthetics, and color affects mood.
That sounds great, radiating a few kilowatts into space on the offchance that a couple of percent of it gets picked up by bulbs and used for lighting.
No. “Networked IoT” normally means cloud connected. No, thanks.
It’s true, that sometimes remote control is nice, but for that a local solution is preferable and enough.
Date those bulbs when new, at least some data can be combed out of brands etc. I find that the 60w equivalent is the cheapest point on the price curve and thermal management is reasonable. I found an eight socket bath fixture that I filled with these type bulbs and it runs cool and if a bulb goes out I am not out the whole fixture like those one piece LED strip light shop-lights at Horror Fright Fools. A friend got some so we’ll see how they go in the long run.
I been working at a hospital for over 13 years as a electrician.
This is what I have noticed.
1 they really only last about twice as long as a incandescent lite.
We have had the fire department down around 3 – 5 times a year because when they burn out some times
they produce a very bad burning smell and people think that there is a fire or a contamination leak.
So the area has to be evacuated. The patients don’t like that or if it is around the Cat scanner that has to be shut down and it can take up to a day to restart them.
Me I think they are OK but in my institute we are probably spending about $5000 to $8000 per call to the Fire department and evacuation and cancellations. The labor and delivery area has all ways been the worst.
But I think if I remember what the hydro bill was, they are still saving money.
We lost big time adding solar heating. It never did work right.And was scrapped only 2 years after installation.
Hydro bill? You must be in East Canada. Solar thermal heating at those latitudes would be a rather poor choice. Such an irony that the latitudes on Earth most suitable for solar thermal space heating need heating the least.
Manitobianites also call it “hydro”.
That is quite Central Canada.
There’s BC Hydro too, so it spreads across the country, though maybe skips some provinces.
I find it really interesting to see such flagrant disregard for quality, coupled with ingeniously cheap design to make something that’s just good enough to sell. Unfortunately low quality products means shorter lifetime and more waste. At least from these products people who wish to design and manufacture near the other end of the spectrum, i.e high quality products, can learn what not to do!
I would encourage anyone who can reasonably afford it not to purchase such cheap products. As fellow designers, engineers, hackers, makers, inventors – let’s together bring back the value of finely crafted products!
How many mice does it take to screw in a light bulb?
Two, but how did they get into the light bulb?
I recently bought a 12 pack of LED bulbs from WalMart for $9.97. Only 60W equivalent but they’ll go into fixtures of less use, and one went into my motion sensing outdoor light to replace a dying halogen. Hello, brightness, my old friend. I can see the stairs again.
My kitchen has a pair of independently switched 4 foot dual T12 fluorescent tube fixtures. For $8.88 each at WalMart I got two pairs of 4 foot T8 LED tubes with universal ballast compatibility. *Must use to replace ALL tubes in a fixture*. Brighter than the old 32W T12 tubes and I got warm white so I like the lighting better in the kitchen. So for $17.76 my kitchen’s contribution to the power bill goes down. I’ll take it.
Now the only non-LED lights in the house are a pair of Circline fixtures, one in the middile of the livingroom ceiling and one in the middle of the master bedroom ceiling, and two ceiling fixtures on the back bedroom ceiling. One takes two candelabra base ‘pointy’ bulbs, the other takes three candelabra base that are a small globe style.
I just counted and there are 54 four foot tubes in my shop. Times 40W = 2160
There are plenty of candelabra style led bulbs at 150, 400 or 500 lumens starting at barely a buck at Lowe’s or home Depot.
Gregg, you’re short on your shop wattage by about 8 – 10 watts per fixture. If they are electronic ballasts it would be lower. Assuming 27 fixtures, that’s about 2,430 watts.
If you want information on how to change them out to LED, at about 11.5 watts each bulb (23 watts total per fixture) I can help you. Reducing it to about 621 watts for your shop. Also when I was replacing them commercially we would reduce them to single bulbs, saving even more. The ability to do that depends greatly on the space and design of the shop though.
I am not selling anything.
I use a lot of the 9W LED bulbs from the dollar tree. I order them by the case. They have replaced all of the old 4′ fluorescent shop lights I used to have in my various shops, and I can not say enough good things about them. They are bright, they are much more rugged than the old tubes, and if you break one it still works, and you do not have a glass mess to clean up. They come up to full brightness right away – regardless of temperature. They do not flicker and hum when it is cold out, or when they have a bug up their ass. And they have a lifetime guarantee. If one dies, toss it in the car and next time you are in town swap it out for a new one. You already have a bucket in your car for the stuff going back to HF for the same reason no doubt. In the house I have replaced all the CFL’s in areas that are not lit for long, like closets, the pantry, over the stairs etc. I still have some CFL’s in places that we keep lit most of the time. I am used to their color and have had them for many years now. The funny thing with the LED’s is that for me at least I have not reaped any power savings. In fact my typical swap out is one fluorescent fixture with 2 40W lamps for one piece of furring strip with 10 9W LED’s on it, costing me an extra 10W, but giving me much more light, and the rest of the benefits I opened this blurb with. So as far as I am concerned as long as I have a spare to two, or ten, these last forever. When one craps out, I just return it. Except for the first one, which of course I dissected. Actually I have dissected a few and in my case, every single one has died from an LED opening up. I made up a little pointy probe to pointy probe jumper and I go through the LED’s shorting each one out and I have not found one that has not had a defective LED. In one case I had one that was a bad solder joint on the LED. I hit one side with the probe and the light came on. Just poking it was enough to push it back into it’s pad. I personally think they are a major step forward for a lot of places.
I started using LED bulbs a couple of years ago. They all go bad after about a month of service. I had one go bad simply because I turned on a box fan. We may have bad power here in Arlington, Texas as our power goes out about once every two weeks. The outages are just long enough for computers to reset (less than a second).
When I moved into my current place. I was burning up these weird shaped tubes, ones that look like the Norse Rune for increased profit, that were $30 a pop, every few months. Getting rapidly tired of this, I got another fixture that took normal bulbs, and as I took the old one out, I realised it had been arcing badly for years. I’ve also seen it from a fractionally popped breaker, snap them all off and back on again once in a while.
Heat is THE one, most efficient destroyer of anything electrical or electronic.
After having corrected the problem of short lifetimes of incandescent bulbs many times in the “pre-green” era by simply finding ways to increase airflow over the bulb(s), I decided to determine why my CFLs weren’t meeting their
statedclaimed lifetimes of 10,000 hours, or more.Post-mortems revealed, almost beyond doubt, that the tremendous amount of heat at the base of a working unit was killing the internal electronics–as evidenced by brown or black components (I’m almost certain that this is what killed those few units with darkened ends, which in earlier times would have been attributed to low starting voltage; or many, many on-off cycles, of conventional fluorescent tubes. Never forget that the long, stated lifetime of a fluorescent tube applies ONLY to a device which is continually on).
I have since restricted my use of CFLs to outdoor lighting which is well cooled (has GOOD airflow over the bulb), stays on continually, and have had some running for as much as two years–17K hours, or so.. Being a pack-rat, and a cheapskate, I laid in a good supply of CFLs when they became exceedingly cheap, so I’ve got years to go for my outdoor lighting.
As far as the now-prevalent LEDs which don’t have the “always-on’ restriction, the only advice is to keep them as cool as possible–do NOT use them in totally enclosed lighting fixtures and expect long lifetimes. And always, ALWAYS avoid the very cheap s**t. It’s (they’re) VERY cheap for a VERY good reason.
My LED bulbs usually die with the bulb flashing. Sometimes they just won’t turn on though. When a LED bulb goes south, is it then the driver which doesn’t work anymore? And if so, what components usually fail? I guess the LEDs themselves usually outlives the electronics, but can a bulb continue working with just one or two dead LEDs?
LED bulbs suck on so many levels. they are expensive, they are VERY faulty (they fail more and more unpredictably that filament bulbs, believe it or not) and 100,000 hours? LOL what a lie.
i have some bulbs that somehow survived beyond 5-6000 hrs (they are usually the very low power ones) and their luminosity is 50 to 20% of what was when new. power consumption however is the same. a really good deal…
all bulbs with 1000plus lumen i bought, their life span was measured in days. and i am not talking about dollar store chinesium stuff here.
how bad it must be when even CFL become an attractive solution?
bottom liine, i ended up building my own light fixtures as any commercial solution sucks donkey or is stupidly expensive.
LED COBs with large heat sinks work decently and i finally can go around the house without a torchlight in the pocket.
Where do you get your LED’s from? I am in a buck a pop for mine at the dollar tree, and they have a lifetime guarantee. I have taken a couple apart to see what the guts look like in them but over the few years I have been using them now I have only had a couple of failures and I am not at all nice to them. The CFL’s were better than the 4′ tubes, but still do not come up to full brightness right away and have some issues when it is really cold out, and are still fragile. That is one of the big ones for me in the shop. I can whack one of the LED’s with a large piece of stock by accident and not spend the next 15 minutes doing a clean up.
There is no trust in the market (in few commodity markets) these days.
The price you pay does not guarantee quality
The brand you buy doesn’t either – OEM, selling on bankrupt company brands
People who care enough rely on teardown reviewers on youtube as media, magazines, websites etc cannot be trusted either as they are looking for advetising revenue so dont always tell the truth either.
Those left over dont care, so just buy the cheapest.
Hence the race to the bottom in anything which is a commodity.
This is all really the fault of the throw away society.
Convenience.
A failure to “make do and mend”
A need to have the latest and new which plagues the majority of the population.
@ Gian, 12:27 am; 18 Feb:
What a tremendously insightful, concise, and profound answer to the entire question concerning LED lamps:
“…how bad it must be when even CFL become an attractive solution?…”
You are 100% correct, Gian.
I’m going to be continue using CFLs as long as I can. Many thanks for your brilliant (no pun intended) observation.
{the only problem I have with higher-power LED COBs, suitable for general-purpose lighting, is that I can’t find any which have a color-temperature in the “normal” range of incandescents: ≈ 4000K; and I absolutely detest “bright white”– ≈6000K–(which have now been re-named “cool-white”, to try and make them more acceptable, I suppose) as a lighting device. Information on any “warm-white” high-power LED COB you may have would be deeply appreciated. “Bright (“Cool”)-White” LEDs are suitable only, marginally, for use in flashlights (‘torches’, to you folks of the British persuasion). Even in those applications, ‘warm-white’ is much more acceptable, as objects tend to be rendered much more ‘naturally’ }.
Have you tried an LED bulb manufactured in this decade? I had a friend who got some of the first LED bulbs, and, yes, they sucked badly. You can get the modern ones in just about any color temperature you like, they take about half the energy of an equivalent CFL, and are much more rugged. I have not tossed all my CFL’s out as some of them are in places that they are suited for, but when they die they will be replaced with LED’s.